# Glossary

Definitions come from the card schema and the tag vocabulary.


## License status (license_status)

What the sources state about the data license.

- **Permissive** `permissive`: All statements agree and allow commercial use and redistribution.
- **Restricted** `restricted`: Any statement prohibits commercial use or derivatives, limits to research, assigns copyright to a third party, or requires payment.
- **Conflict** `conflict`: Statements contradict each other.
- **Unstated** `unstated`: No statement.
- **Unverifiable** `unverifiable`: Page could not be opened or only second-hand information.

## Mdexa hosting (hosting)

What Mdexa does with the dataset, separate from the license.

- **Hosted** `hosted`: Files served on Mdexa (requires permissive, no risk flags, and signoff).
- **Eligible** `eligible`: Permissive and no risk flags, awaiting signoff.
- **Link only** `link-only`: Restricted, or permissive with any risk flag.
- **Pending** `pending`: License_status conflict, unstated or unverifiable.

## Source risk flags (risk_flags)

Risks that a license verdict does not cover.

- **Exam-derived** `exam-derived`: Questions come from copyrighted examinations.
- **Scraped** `scraped`: Content was collected from websites.
- **Real patient text** `real-patient-text`: Derived from real patient text; privacy authorisation needs checking.
- **Model-generated** `model-generated`: Generated by a model; the model's output terms need checking.
- **Derived** `derived`: Processed from other datasets, whose terms also apply.
- **Unknown provenance** `unknown-provenance`: The source is not stated, so its copyright cannot be judged.

## Task format (task_format)

- **Multiple-choice (single answer)** `choice-question`: Items with options and one correct answer.
- **Multiple-answer choice** `multi-answer`: Options with more than one correct answer allowed.
- **True/false** `true-false`: Answered true or false.
- **Fill-in-the-blank** `fill-in-blank`: Short answer filling a gap.
- **Open-ended QA** `open-qa`: Single-turn question with free-text answer, incl. instructions.
- **Multi-turn dialogue** `multi-turn-dialogue`: Conversations with several turns.
- **Named entity recognition** `ner`: Entity spans and types annotated in text.
- **Relation extraction** `relation-extraction`: Relations between entities annotated in text.
- **Term normalization** `term-normalization`: Map mentions to standard terms or codes.
- **Text classification** `text-classification`: Assign a label to a text or utterance.
- **Text matching** `text-matching`: Similarity or relevance between texts, incl. answer selection.
- **Summarization / report generation** `summarization`: Produce a summary or report from source text.
- **Visual QA** `visual-qa`: Questions that require reading an image.

## Intended use (intended_use)

- **Pretraining** `pretraining`: Raw text intended for (continued) pretraining.
- **Fine-tuning** `fine-tuning`: Supervised or instruction fine-tuning data.
- **Evaluation** `evaluation`: Intended as a benchmark or test set.
- **Retrieval knowledge base** `retrieval-kb`: Intended as a corpus or KB to retrieve from.

## Specialty (specialty)

- **Internal medicine (general)** `internal-medicine`: Datasets explicitly limited to internal medicine (general).
- **Cardiology** `cardiology`: Datasets explicitly limited to cardiology.
- **Respiratory medicine** `respiratory`: Datasets explicitly limited to respiratory medicine.
- **Gastroenterology** `gastroenterology`: Datasets explicitly limited to gastroenterology.
- **Endocrinology** `endocrinology`: Datasets explicitly limited to endocrinology.
- **Nephrology** `nephrology`: Datasets explicitly limited to nephrology.
- **Hematology** `hematology`: Datasets explicitly limited to hematology.
- **Neurology** `neurology`: Datasets explicitly limited to neurology.
- **Rheumatology and immunology** `rheumatology-immunology`: Datasets explicitly limited to rheumatology and immunology.
- **Infectious diseases** `infectious-diseases`: Datasets explicitly limited to infectious diseases.
- **Surgery** `general-surgery`: Datasets explicitly limited to surgery.
- **Orthopedics** `orthopedics`: Datasets explicitly limited to orthopedics.
- **Urology and andrology** `urology-andrology`: Datasets explicitly limited to urology and andrology.
- **Obstetrics and gynecology** `obstetrics-gynecology`: Datasets explicitly limited to obstetrics and gynecology.
- **Pediatrics** `pediatrics`: Datasets explicitly limited to pediatrics.
- **Ophthalmology** `ophthalmology`: Datasets explicitly limited to ophthalmology.
- **Otorhinolaryngology** `otorhinolaryngology`: Datasets explicitly limited to otorhinolaryngology.
- **Stomatology and dentistry** `stomatology`: Datasets explicitly limited to stomatology and dentistry.
- **Dermatology** `dermatology`: Datasets explicitly limited to dermatology.
- **Psychiatry** `psychiatry`: Datasets explicitly limited to psychiatry.
- **Oncology** `oncology`: Datasets explicitly limited to oncology.
- **Emergency and critical care** `emergency-critical-care`: Datasets explicitly limited to emergency and critical care.
- **Anesthesiology** `anesthesiology`: Datasets explicitly limited to anesthesiology.
- **Rehabilitation medicine** `rehabilitation`: Datasets explicitly limited to rehabilitation medicine.
- **General practice and basic medical sciences** `general-practice-basic`: Datasets explicitly limited to general practice and basic medical sciences.
- **Medical imaging** `medical-imaging`: Datasets explicitly limited to medical imaging.
- **Laboratory medicine and pathology** `laboratory-pathology`: Datasets explicitly limited to laboratory medicine and pathology.
- **Pharmacy** `pharmacy`: Datasets explicitly limited to pharmacy.
- **Nursing** `nursing`: Datasets explicitly limited to nursing.
- **Public health and preventive medicine** `public-health-preventive`: Datasets explicitly limited to public health and preventive medicine.
- **TCM internal medicine** `tcm-internal`: Datasets explicitly limited to tcm internal medicine.
- **TCM surgery** `tcm-surgery`: Datasets explicitly limited to tcm surgery.
- **TCM gynecology** `tcm-gynecology`: Datasets explicitly limited to tcm gynecology.
- **TCM pediatrics** `tcm-pediatrics`: Datasets explicitly limited to tcm pediatrics.
- **Acupuncture and tuina** `acupuncture-tuina`: Datasets explicitly limited to acupuncture and tuina.
- **TCM orthopedics and traumatology** `tcm-orthopedics-traumatology`: Datasets explicitly limited to tcm orthopedics and traumatology.
- **Chinese materia medica and formulas** `tcm-materia-formulas`: Datasets explicitly limited to chinese materia medica and formulas.
- **TCM fundamentals and classics** `tcm-basics-classics`: Datasets explicitly limited to tcm fundamentals and classics.
- **Multi-specialty** `multi-specialty`: Datasets explicitly limited to multi-specialty.

## Disease classification (ICD-11 chapter) (condition_scope)

- **Certain infectious or parasitic diseases** `icd11-01`: Restricted to ICD-11 chapter 01.
- **Neoplasms** `icd11-02`: Restricted to ICD-11 chapter 02.
- **Diseases of the blood or blood-forming organs** `icd11-03`: Restricted to ICD-11 chapter 03.
- **Diseases of the immune system** `icd11-04`: Restricted to ICD-11 chapter 04.
- **Endocrine, nutritional or metabolic diseases** `icd11-05`: Restricted to ICD-11 chapter 05.
- **Mental, behavioural or neurodevelopmental disorders** `icd11-06`: Restricted to ICD-11 chapter 06.
- **Sleep-wake disorders** `icd11-07`: Restricted to ICD-11 chapter 07.
- **Diseases of the nervous system** `icd11-08`: Restricted to ICD-11 chapter 08.
- **Diseases of the visual system** `icd11-09`: Restricted to ICD-11 chapter 09.
- **Diseases of the ear or mastoid process** `icd11-10`: Restricted to ICD-11 chapter 10.
- **Diseases of the circulatory system** `icd11-11`: Restricted to ICD-11 chapter 11.
- **Diseases of the respiratory system** `icd11-12`: Restricted to ICD-11 chapter 12.
- **Diseases of the digestive system** `icd11-13`: Restricted to ICD-11 chapter 13.
- **Diseases of the skin** `icd11-14`: Restricted to ICD-11 chapter 14.
- **Diseases of the musculoskeletal system or connective tissue** `icd11-15`: Restricted to ICD-11 chapter 15.
- **Diseases of the genitourinary system** `icd11-16`: Restricted to ICD-11 chapter 16.
- **Conditions related to sexual health** `icd11-17`: Restricted to ICD-11 chapter 17.
- **Pregnancy, childbirth or the puerperium** `icd11-18`: Restricted to ICD-11 chapter 18.
- **Certain conditions originating in the perinatal period** `icd11-19`: Restricted to ICD-11 chapter 19.
- **Developmental anomalies** `icd11-20`: Restricted to ICD-11 chapter 20.
- **Symptoms, signs or clinical findings, not elsewhere classified** `icd11-21`: Restricted to ICD-11 chapter 21.
- **Injury, poisoning or certain other consequences of external causes** `icd11-22`: Restricted to ICD-11 chapter 22.
- **External causes of morbidity or mortality** `icd11-23`: Restricted to ICD-11 chapter 23.
- **Factors influencing health status or contact with health services** `icd11-24`: Restricted to ICD-11 chapter 24.
- **Codes for special purposes** `icd11-25`: Restricted to ICD-11 chapter 25.
- **Supplementary Chapter Traditional Medicine Conditions - Module 1** `icd11-26`: Restricted to ICD-11 chapter 26.

## Source type (source_type)

- **Exam questions** `exam`: Licensing or entrance exam questions. Derived from risk_flags; never stored.
- **Textbooks and reference books** `textbook-reference`: Compiled from textbooks or reference books.
- **Clinical guidelines** `guideline`: From clinical guidelines or consensus.
- **Research literature** `literature`: From papers or journal articles.
- **Expert-written** `expert-written`: Newly written by clinicians or experts.
- **Built from a knowledge base** `knowledge-base`: Built from a structured KB or KG.
- **Classical texts** `classical-text`: From pre-modern classical texts.
- **Official publication** `official-publication`: Published by a government or standards body.
- **Real patient text** `real-patient-text`: From real consultations or records. Derived from risk_flags; never stored.
- **Model-generated** `model-generated`: Generated by a model. Derived from risk_flags; never stored.
- **Web-scraped** `web-scraped`: Scraped from websites. Derived from risk_flags (scraped); never stored.
- **Derived from other datasets** `derived-dataset`: Built from other datasets. Derived from risk_flags (derived); never stored.

## Annotation depth (annotation_depth)

- **With explanations** `with-explanation`: Reference explanations or rationales included.
- **With reasoning traces** `with-reasoning`: Step-by-step reasoning included.
- **With evidence citations** `with-evidence`: Answers cite source passages.

## Expert involvement (expert_involvement)

- **None** `none`: Explicitly no expert review.
- **Reviewed, method unspecified** `review-unspecified`: Experts reviewed; method not stated.
- **Sample review** `sample-review`: Experts reviewed a sample.
- **Per-item review** `per-item-review`: Every item (or a stated subset) reviewed.
- **Multiple independent reviewers** `multi-reviewer`: Independent reviews with agreement recorded.

## Data format (data_format)

- **JSON** `json`: Data files provided as json.
- **JSONL** `jsonl`: Data files provided as jsonl.
- **CSV** `csv`: Data files provided as csv.
- **TSV** `tsv`: Data files provided as tsv.
- **PARQUET** `parquet`: Data files provided as parquet.
- **TXT** `txt`: Data files provided as txt.
- **XLSX** `xlsx`: Data files provided as xlsx.

## Official split (official_split)

- **Train and test splits** `full-split`: Official training and test splits.
- **Evaluation split only** `eval-only`: Only test or validation splits.
- **No split** `no-split`: Single undivided set.

## Benchmark status (benchmark_status)

- **Public leaderboard** `public-leaderboard`: An official public leaderboard exists.
- **Human expert baseline** `human-expert-baseline`: Expert human performance reported.